AI has learnt to authenticate sneakers by their smell

Artificial intelligence has learnt to recognise real sneakers from fake onesThe growing trend of dupes in fashion has motivated several entrepreneurs and inventors to invest in tools capable of certifying the authenticity of items. Osmo, a startup co-founded by a former Google researcher, Alex Wiltschko, focuses on exactly that. The company is not only specialized in recognizing authentic sneakers, but in all scents. Osmo is indeed able to create and recognize new fragrances as well as to detect diseases in patients (similar to trained dogs, one might say). Thanks to a computerized recognition system, Osmo's technology has digitized the sense of smell. This invention could forever innovate the retail industry, speeding up the process of item verification and offering more protection to businesses currently struggling due to the success of the counterfeit market.

Osmo began to take shape in 2022, when Wiltschko was still working as a researcher for Google Research. By providing artificial intelligence with 5,000 different aromatic molecules, the co-founder’s team was able to create a true “scent map”, capable of categorizing each fragrance - or odor - in a structured way. Since there was no online glossary of scents to use as a basis for the map, in addition to the algorithm and AI, the team behind Osmo collaborated with master perfumers who could describe scents in fine detail. The explanation behind the certification process for authentic sneakers is quite simple: because counterfeit shoe manufacturers tend to use cheaper materials of lower quality, the scent of a dupe differs from that of an original. According to the company, recent tests have shown that Osmo's accuracy in identifying counterfeits currently reaches 95%.

Currently, Osmo is able to certify sneakers and bags, items whose materials are easier to identify since they are typically made of leather or suede (while dupes often use synthetic or lower-quality leathers). It will take time for the startup's AI to learn to distinguish genuine designer clothes from counterfeits, as their scents are less intense and thus harder for Osmo’s computers to interpret. The company now faces several obstacles: the first concerns the “recruitment” process for a sneaker in the AI database, which could require up to fifty pairs of shoes – a quantity that may be unavailable for limited drops. Another challenge is the future of this tool in the hands of retailers, as the AI for scent recognition requires significant amounts of time and scientific knowledge for maintenance and updates. Nevertheless, Wiltschko's startup innovation once again highlights the fashion industry's growing interest in item certification, a topic of increasing importance for major luxury brands and online resale platforms that want to protect their products and, in turn, their reputation.